Research Summary: Automated identification of ivory carvings using machine learning: A proof-of-concept approach to detecting the illegal wildlife trade

A research study introduces a proof-of-concept machine learning approach to automate the detection of illegal elephant ivory carvings in online marketplaces. The work addresses severe operational bottlenecks in wildlife law enforcement, where high advert volumes, visual resemblance to legal materials, and the physical constraints of traditional identification techniques impede proactive monitoring and prosecution.

Why it matters

This initiative addresses a foundational enforcement challenge by seeking to automate the visual identification of illicit wildlife products across digital channels. Bridging the operational gap between physical lab-based diagnostics and high-volume digital marketplace screening is critical for scaling enforcement without exponential increases in manual labor. Establishing automated detection capabilities could fundamentally alter how regulatory compliance and cross-border trade controls are operationalised online.

Key insights

• International bans have failed to eliminate illicit trade, with poached elephant ivory continuing to be traded actively across digital marketplaces.
• Enforcement and prosecution remain hindered by massive advert volumes and the visual similarity between ivory and legal alternative materials.
• Standard diagnostic techniques such as Fourier-transform infrared spectroscopy and optical microscopy require physical item access, making them unviable for screening online photographs.
• The source presents a machine learning solution to bridge this gap, although the provided text truncates before detailing model architecture and empirical performance.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2610.33243